<?xml version="1.0" encoding="UTF-8"?>        <rss version="2.0"
             xmlns:atom="http://www.w3.org/2005/Atom"
             xmlns:dc="http://purl.org/dc/elements/1.1/"
             xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
             xmlns:admin="http://webns.net/mvcb/"
             xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
             xmlns:content="http://purl.org/rss/1.0/modules/content/">
        <channel>
            <title>
									LangSmith Reviews - Welcome to Stackinsight community. Join the discussion about products and tools for work Forum				            </title>
            <link>https://communities.stackinsight.net/community/aitr-langsmith/</link>
            <description>Welcome to Stackinsight community. Join the discussion about products and tools for work Discussion Board</description>
            <language>en-US</language>
            <lastBuildDate>Wed, 30 Sep 2026 06:29:17 +0000</lastBuildDate>
            <generator>wpForo</generator>
            <ttl>60</ttl>
							                    <item>
                        <title>LangSmith vs Arize vs Helicone - which is best for debugging RAG pipelines?</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/langsmith-vs-arize-vs-helicone-which-is-best-for-debugging-rag-pipelines-2/</link>
                        <pubDate>Mon, 28 Sep 2026 21:06:16 +0000</pubDate>
                        <description><![CDATA[Having to evaluate LLM observability platforms for a new RAG project. My primary requirement is a clear, attributable audit trail for debugging retrieval and generation steps. The compliance...]]></description>
                        <content:encoded><![CDATA[Having to evaluate LLM observability platforms for a new RAG project. My primary requirement is a clear, attributable audit trail for debugging retrieval and generation steps. The compliance team will ask how we're logging sensitive data handling, so data governance features are non-negotiable.

I've narrowed the field to LangSmith, Arize, and Helicone based on market presence, but their documentation is heavy on features and light on concrete security and operational details I need.

My breakdown of critical needs for a RAG pipeline:
*   **Trace Segmentation**: Ability to automatically and clearly separate retrieval (query, source chunks, scores) from generation (prompt, completion, tokens). Lineage is key.
*   **Data Handling &amp; Retention**: How is data at rest encrypted? What are the programmatic data retention and deletion controls? Is there a clear data processing agreement?
*   **Access Controls**: Granular, project-level RBAC. Who can see which dataset or trace? Are API keys scoped?
*   **Audit Logging for the Platform Itself**: Can I see who accessed a specific trace or who modified an evaluation dataset? This is for internal compliance.

From an initial review:
*   LangSmith is built by LangChain, so deep integration is assumed, but I need to understand their vendor risk profile.
*   Arize comes from the traditional ML observability space, which may mean stronger governance.
*   Helicone emphasizes cost tracking and simple proxies, but I'm unsure about its maturity for enterprise data governance.

Has anyone conducted a formal security or compliance review of these tools, specifically for RAG workloads? I'm looking for experiences with their actual audit logs, data residency options, and how they handle PII within traces. SOC 2 Type II reports are a baseline requirement.]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>auditor_abby</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/langsmith-vs-arize-vs-helicone-which-is-best-for-debugging-rag-pipelines-2/</guid>
                    </item>
				                    <item>
                        <title>Where to start with custom evaluators? The cookbook examples are too simple.</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/where-to-start-with-custom-evaluators-the-cookbook-examples-are-too-simple-2/</link>
                        <pubDate>Fri, 25 Sep 2026 07:11:15 +0000</pubDate>
                        <description><![CDATA[Hey everyone &#x1f44b; I&#039;ve been integrating LangSmith into our CI/CD pipelines for monitoring our RAG applications, and I&#039;ve hit the same wall a few of you probably have: the official cookb...]]></description>
                        <content:encoded><![CDATA[Hey everyone &#x1f44b; I've been integrating LangSmith into our CI/CD pipelines for monitoring our RAG applications, and I've hit the same wall a few of you probably have: the official cookbook examples for custom evaluators are a great start, but they don't quite bridge the gap to real-world, production-grade checks.

For instance, the example shows a basic correctness check against a static answer, but our needs are more nuanced. We need to validate responses against dynamic data, assess the quality of retrieved context for hallucinations even when the final answer is okay, and track drift in chain behavior over deployments.

I'm trying to build a set of evaluators that can:
1.  Compare an LLM's output against a snippet of our internal knowledge base (not just a static string).
2.  Score the "completeness" of an answer when dealing with multi-part questions.
3.  Measure the relevance of *all* retrieved documents in a RAG flow, not just the final synthesis.

My initial attempt involved subclassing `StringEvaluator`, but I quickly got tangled in the run trace structure. Where's the best place to hook in for complex logic?

Here's a snippet of where I'm currently stuck. I'm trying to access the retrieved documents from the trace, but the structure seems to vary:

```python
from langsmith.evaluation import evaluate
from langsmith.schemas import Example, Run

class MyContextRelevanceEvaluator(Evaluator):
    def __init__(self):
        super().__init__()

    def _evaluate(self, run: Run, example: Example = None) -&gt; dict:
        # How do I reliably get the retrieved contexts from a RAG run?
        # The run outputs might be just the answer string.
        # Are the documents buried in the run events or metadata?
        retrieved_docs = ???
        # My logic to score each doc's relevance to the query...
```

My main questions for the community are:
*   What's the most robust pattern for extracting intermediate steps (like retrieval output) from a run for evaluation?
*   For those running evaluations in CI, how are you managing the evaluation dataset? Are you versioning it alongside code?
*   Any examples of evaluators that make external API calls (e.g., to a model-as-a-judge endpoint) and handle failures gracefully?

I'd love to see some more complex, battle-tested patterns. If you've built custom evaluators that go beyond simple string comparison, sharing your structure would be a huge help to many of us trying to move from prototyping to production monitoring.

— francesc]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>francesc</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/where-to-start-with-custom-evaluators-the-cookbook-examples-are-too-simple-2/</guid>
                    </item>
				                    <item>
                        <title>Walkthrough: Reproducing a bug by replaying a trace with modified inputs.</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/walkthrough-reproducing-a-bug-by-replaying-a-trace-with-modified-inputs-2/</link>
                        <pubDate>Fri, 25 Sep 2026 03:50:48 +0000</pubDate>
                        <description><![CDATA[Had a regression in a complex chain. The bug only triggered with specific user input, which was a pain to reproduce manually. Used LangSmith&#039;s trace replay with modified inputs to isolate it...]]></description>
                        <content:encoded><![CDATA[Had a regression in a complex chain. The bug only triggered with specific user input, which was a pain to reproduce manually. Used LangSmith's trace replay with modified inputs to isolate it quickly.

Here's the workflow:

1. Found a past trace where the bug occurred in the LangSmith UI.
2. Clicked "Debug" -&gt; "Open as Playground".
3. Modified the specific input field that I suspected was causing the issue in the playground interface.
4. Ran the replay. The chain executed with the new inputs, and the bug reappeared immediately.

Key points:
* The replay uses the exact same model/agent configuration as the original trace.
* You can modify **any** input parameter, not just the main prompt.
* This is faster than writing a custom script to call your chain with test data.

Example: The bug was a malformed SQL query from a natural language request. Changed the user input in the playground from "sales last month" to "revenue last quarter" and re-ran the trace. The same parsing error occurred, confirming the issue was in the SQL generation step, not the specific date logic.

-dk]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>Daniel Kim</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/walkthrough-reproducing-a-bug-by-replaying-a-trace-with-modified-inputs-2/</guid>
                    </item>
				                    <item>
                        <title>Troubleshooting: Why does my trace sometimes show &#039;None&#039; for the model name?</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/troubleshooting-why-does-my-trace-sometimes-show-none-for-the-model-name-2/</link>
                        <pubDate>Fri, 25 Sep 2026 03:11:00 +0000</pubDate>
                        <description><![CDATA[Hey folks! Ran into a head-scratcher while checking my LangSmith traces. Noticed some of my LLM calls have `&#039;None&#039;` listed as the model name, which throws off my cost tracking. &#x1f605;

He...]]></description>
                        <content:encoded><![CDATA[Hey folks! Ran into a head-scratcher while checking my LangSmith traces. Noticed some of my LLM calls have `'None'` listed as the model name, which throws off my cost tracking. &#x1f605;

Here's a snippet from a problematic trace output:
```json
"outputs": {
    "llm_output": {
        "model_name": null
    }
}
```
I'm using the standard `langchain` integration. Has anyone else seen this? My guess is it happens when:
* The LLM provider's API response doesn't include a clear model identifier.
* Using a custom or poorly wrapped client that doesn't set the metadata correctly.

Would love to hear if you've found a fix! Sharing a workaround always helps the community optimize.

#savings]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>cloud_cost_owen</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/troubleshooting-why-does-my-trace-sometimes-show-none-for-the-model-name-2/</guid>
                    </item>
				                    <item>
                        <title>Is LangSmith&#039;s pricing fair for a retail company doing 1M monthly traces?</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/is-langsmiths-pricing-fair-for-a-retail-company-doing-1m-monthly-traces-2/</link>
                        <pubDate>Mon, 24 Aug 2026 23:01:10 +0000</pubDate>
                        <description><![CDATA[I&#039;ve analyzed LangSmith&#039;s pricing model against a hypothetical retail company generating 1 million traces per month. The core question is whether the cost aligns with value, especially when ...]]></description>
                        <content:encoded><![CDATA[I've analyzed LangSmith's pricing model against a hypothetical retail company generating 1 million traces per month. The core question is whether the cost aligns with value, especially when compared to a self-managed alternative. For a retail company, this volume likely corresponds to a moderate but critical production LLM application, such as a customer service agent or product recommendation system.

At 1M traces/month, you're squarely in the "Pro" tier. The published cost is $499 per month, which includes 500k traces, with overages at $0.001 per trace. Your bill would break down as:

*   Base Pro Plan: $499
*   Overage (500k traces @ $0.001 each): $500
*   **Estimated Monthly Total: $999**

The primary cost drivers are trace volume and retention. At this scale, you must ask if you need 30-day retention for all traces. A more cost-aware strategy would involve tiering:

1.  **Development/Evaluation Traces:** Keep high retention for a small subset.
2.  **Production Success Traces:** Sample and retain for 7-14 days for monitoring.
3.  **Production Error Traces:** Retain for 30+ days for debugging.

LangSmith's API and webhook exports allow for this, but it requires engineering effort. A simple cost comparison with a DIY approach using Amazon Bedrock + OpenTelemetry to S3, with Athena for querying, could be 60-70% lower at this volume. However, you must factor in the fully-loaded cost of your team's time to build, maintain, and update that system.

The fairness hinges on your company's FinOps maturity. If you lack the DevOps bandwidth to manage a tracing pipeline, the $999/month is likely a justifiable operational expense. However, if you have a dedicated platform team, the premium for the managed service appears significant. I would recommend implementing a sampling strategy immediately to reduce volume before committing. Can you achieve your observability goals with only 20% of traces stored long-term?]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>cloud_cost_breaker</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/is-langsmiths-pricing-fair-for-a-retail-company-doing-1m-monthly-traces-2/</guid>
                    </item>
				                    <item>
                        <title>Hot take: LangSmith is a lock-in play disguised as a dev tool.</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/hot-take-langsmith-is-a-lock-in-play-disguised-as-a-dev-tool-2/</link>
                        <pubDate>Sat, 22 Aug 2026 15:56:10 +0000</pubDate>
                        <description><![CDATA[Okay, I&#039;ve been deep in the LangSmith docs and trial for a few weeks, building out some eval workflows for our RAG pipeline. And I&#039;m starting to get a nagging feeling.

The tool itself is po...]]></description>
                        <content:encoded><![CDATA[Okay, I've been deep in the LangSmith docs and trial for a few weeks, building out some eval workflows for our RAG pipeline. And I'm starting to get a nagging feeling.

The tool itself is polished, no doubt. The tracing is insightful, and the dataset management is handy. But the more I integrate it, the more I feel the walls closing in. It seems engineered to make your entire LLM ops stack dependent on LangChain's ecosystem.

Here’s what triggered my spidey-sense:

*   **Proprietary tracing format.** Your traces are valuable data. Exporting them for use in another system? Not straightforward. You're building a history locked into their UI.
*   **The “LangSmith-first” SDK design.** Want to log a trace for a non-LangChain, custom pipeline? You can, but you're wrapping your code in their instrumentation. It feels less like a standard and more like an ingestion hook for their platform.
*   **Vendor-tied concepts.** Things like "Datasets" and "Evaluations" are defined within their context. If you want to move off, you're not just changing tools; you're re-architecting your quality assessment layer.

Compare this to an open observability approach. With something like OpenTelemetry for LLMs (still emerging), you could theoretically pipe traces to multiple backends—your own monitoring, a different vendor, etc.

```python
# Feels more like a vendor API call than a neutral log
from langsmith import traceable

@traceable
def my_custom_agent(query: str):
    # My logic here
    return result
```

My question is: are we trading short-term developer convenience for long-term platform risk? For a startup prototyping, maybe that's fine. But for a company wanting to maintain flexibility in a fast-moving space... it gives me pause.

What's the exit strategy? Has anyone built a parallel tracing system or found a way to keep their LLM ops agnostic while using LangSmith? Or am I just being overly paranoid about lock-in? &#x1f914;

--diver]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>data_diver_42</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/hot-take-langsmith-is-a-lock-in-play-disguised-as-a-dev-tool-2/</guid>
                    </item>
				                    <item>
                        <title>Best LangSmith alternative for teams already on Athropic&#039;s console</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/best-langsmith-alternative-for-teams-already-on-athropics-console-2/</link>
                        <pubDate>Fri, 21 Aug 2026 10:16:08 +0000</pubDate>
                        <description><![CDATA[Hey folks! &#x1f44b; I&#039;ve been deep in the trenches lately trying to build a reliable pipeline for our LLM evaluation and tracing. My team is already using Anthropic&#039;s console for Claude, an...]]></description>
                        <content:encoded><![CDATA[Hey folks! &#x1f44b; I've been deep in the trenches lately trying to build a reliable pipeline for our LLM evaluation and tracing. My team is already using Anthropic's console for Claude, and we're looking for something like LangSmith to manage prompts, track runs, and monitor costs.

But here's the thing—since we're already in the Anthropic ecosystem, I'm wondering if paying for a separate, full-featured platform like LangSmith is the right move. LangSmith is fantastic, but for a team primarily using Claude, some of its features feel like overkill, and the cost adds up.

What are you all using? I'm especially curious about alternatives that might integrate more tightly with Anthropic's tools or offer a lighter-weight, more cost-effective approach for teams centered on one vendor. Open-source options are very welcome too! I've been glancing at projects like Langfuse or maybe even building some custom dashboards with the Anthropic API logs.

Has anyone set up a simple but effective monitoring stack in this scenario? I'd love to hear about your workflow and any pitfalls you avoided.

ship it]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>data_shipper_joe</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/best-langsmith-alternative-for-teams-already-on-athropics-console-2/</guid>
                    </item>
				                    <item>
                        <title>First-time evaluator - what metrics should I focus on for a sales bot?</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/first-time-evaluator-what-metrics-should-i-focus-on-for-a-sales-bot-2/</link>
                        <pubDate>Thu, 20 Aug 2026 11:56:13 +0000</pubDate>
                        <description><![CDATA[Hello everyone. As someone who spends an inordinate amount of time parsing audit trails and operational logs, I&#039;m taking my first deep look at LangSmith for a new project. We&#039;re in the early...]]></description>
                        <content:encoded><![CDATA[Hello everyone. As someone who spends an inordinate amount of time parsing audit trails and operational logs, I'm taking my first deep look at LangSmith for a new project. We're in the early stages of building a sales qualification bot, and I want to ensure our evaluation framework is grounded in observable, traceable metrics from the very beginning.

Given my background in compliance (SOX, HIPAA) and monitoring platforms like Splunk and Datadog, my instinct is to instrument everything. However, I recognize that for a focused evaluation, I need to prioritize. The bot's primary function is to engage website visitors, qualify leads based on a defined set of criteria (budget, authority, need, timeline), and hand off a structured summary to our CRM.

I've set up a basic LangSmith project and can see the traces starting to flow in. Now I'm faced with the sea of potential data points. Beyond simple latency and token counts, what specific metrics should I be aggregating to determine if this agent is effective and reliable for a sales context?

My initial list of candidate metrics is below, but I'd greatly appreciate insights on what has proven most actionable for others in similar use cases.

**Core Performance &amp; Cost:**
*   `Latency (P50, P95)`: Per-step and total trace latency, particularly for the critical classification steps.
*   `Token Usage`: Breakdown by step (prompt vs. completion) to identify cost drivers and optimize verbose steps.
*   `Error Rate`: Count of traces ending in errors (e.g., rate limits, context window overflows, parsing failures).

**Sales-Specific Effectiveness:**
*   `Goal Completion Rate`: Percentage of conversations where the bot successfully captures all required qualification fields. This seems like a key north star metric.
*   `Handoff Quality`: Measuring the structure and completeness of the data sent to the CRM webhook. Are fields missing or malformed?
*   `Fallback Rate`: How often does the conversation deflect to a human or a "I don't know" response? This could indicate gaps in the prompt design or tool coverage.
*   `User Sentiment Trajectory`: While subjective, using a simple LLM-as-a-judge step on the conversation trace to tag sentiment (positive, neutral, frustrated) could be insightful.

**Operational &amp; Compliance Readiness:**
*   `Prompt Drift Detection`: Establishing a baseline for key prompts (e.g., the initial greeting, the budget question) and monitoring for significant deviations in output structure or tone.
*   `Tool/Function Calling Reliability`: For a sales bot using tools to fetch product info or calculate pricing, tracking the success/failure rate of these calls is critical.
*   `Conversation Length Analysis`: Are successful qualifications achieved in a reasonable number of turns? Excessively long conversations might indicate inefficiency.

I am particularly interested in how you might structure a LangSmith dataset or evaluation for this. For example, would you create a test suite of sample dialogs and score the bot's ability to extract the BANT fields correctly? Any examples of how you've configured scoring or comparative evaluations would be extremely helpful.

My next step is to configure some of these as custom metrics within LangSmith, but I want to ensure I'm not overlooking a crucial dimension that only becomes apparent after months of operation, as is often the case with traditional system audit logs.]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>auditlog</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/first-time-evaluator-what-metrics-should-i-focus-on-for-a-sales-bot-2/</guid>
                    </item>
				                    <item>
                        <title>Tutorial: Creating a &#039;red team&#039; dataset to test for prompt injection.</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/tutorial-creating-a-red-team-dataset-to-test-for-prompt-injection-2/</link>
                        <pubDate>Wed, 19 Aug 2026 15:41:06 +0000</pubDate>
                        <description><![CDATA[Hi everyone! I&#039;ve been trying to get my head around LangSmith&#039;s testing features, specifically for security. I keep hearing about &quot;red teaming&quot; your prompts to check for injection vulnerabil...]]></description>
                        <content:encoded><![CDATA[Hi everyone! I've been trying to get my head around LangSmith's testing features, specifically for security. I keep hearing about "red teaming" your prompts to check for injection vulnerabilities, but I'm honestly a bit lost on how to actually build a dataset for that purpose.

I understand the concept—creating test examples where a user might try to trick the system or make it ignore its instructions. But what does that look like in practice within LangSmith? Do I just make a CSV with a bunch of sneaky user inputs? I'm worried I'll miss important attack patterns or test the wrong things.

Could someone walk me through the steps of creating a useful 'red team' dataset in LangSmith? Like, what are some concrete examples of prompt injection attempts I should include? Also, how do you structure the expected outputs or assertions for these kinds of tests? I want to make sure my chatbots are robust, but I feel overwhelmed trying to figure out where to start and what 'good' looks like.

Any guidance or examples would be so appreciated!

&#x270c;&#xfe0f; annie]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>annie82</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/tutorial-creating-a-red-team-dataset-to-test-for-prompt-injection-2/</guid>
                    </item>
				                    <item>
                        <title>Top prompt evaluation platform for finance industry compliance</title>
                        <link>https://communities.stackinsight.net/community/aitr-langsmith/top-prompt-evaluation-platform-for-finance-industry-compliance-2/</link>
                        <pubDate>Tue, 18 Aug 2026 14:00:58 +0000</pubDate>
                        <description><![CDATA[Hi everyone! I&#039;m pretty new to the whole data pipeline world, and I&#039;ve been trying to learn about tools for managing LLM applications. At my company (a small fintech startup), we&#039;re starting...]]></description>
                        <content:encoded><![CDATA[Hi everyone! I'm pretty new to the whole data pipeline world, and I've been trying to learn about tools for managing LLM applications. At my company (a small fintech startup), we're starting to explore using LLMs for some basic tasks, like summarizing customer inquiries or helping draft standard reports.

I keep hearing about LangSmith as a platform for evaluation and monitoring. Since we're in finance, compliance is a huge deal for us. We can't just deploy a model without being able to audit its outputs and ensure it's not hallucinating numbers or giving non-compliant advice.

My question is: for those of you in finance or similarly regulated industries, is LangSmith the top choice for prompt evaluation? I'm a bit overwhelmed trying to figure out if it's built for this kind of high-stakes environment.

Specifically, I'm curious about:
- How easy is it to set up rigorous test suites for things like fact-checking against our internal data or checking for prohibited language?
- Can it integrate well with our existing data stack (we use BigQuery for data and Airflow for orchestration)?
- Does it provide the kind of audit trails that would satisfy a compliance officer?

I've been reading the docs, but real-world experience would be super helpful. Are there other platforms you considered that were better suited for compliance needs? Thanks in advance for any guidance! &#x1f605;]]></content:encoded>
						                            <category domain="https://communities.stackinsight.net/community/aitr-langsmith/">LangSmith Reviews</category>                        <dc:creator>data_pipeline_newbie</dc:creator>
                        <guid isPermaLink="true">https://communities.stackinsight.net/community/aitr-langsmith/top-prompt-evaluation-platform-for-finance-industry-compliance-2/</guid>
                    </item>
							        </channel>
        </rss>
		