Botpress
A reliable knowledge-base chatbot for policy Q&A that keeps context well and handles sensitive messages responsibly.
Accurate retrieval with a few production-readiness gaps
Botpress answered all three policy scenarios accurately, retained context well across follow-up questions, and handled a self-harm statement with appropriate empathy instead of forcing it back into a support flow. The main issue in this test was a confusing follow-up on EMI reversals: after correctly saying customized laptop orders are not refundable after payment confirmation, it still gave a refund-timeline answer that could mislead a customer unless exceptional-case approval is clarified. The report also did not document visible citations or source attribution, so its RAG grounding looked strong in practice but was not fully verified against that part of the use case.
In-Depth Review
Our detailed analysis of Botpress — features, performance, and real-world testing.
Feature-by-Feature Breakdown
Knowledge-grounded policy answeringHandled direct policy questions accurately across returns, delivery compensation, and cancellation rules, with clear conditions and exclusions.▾
Feature tested: Knowledge-grounded policy answering
Result: Passed
Verdict: Handled direct policy questions accurately across returns, delivery compensation, and cancellation rules, with clear conditions and exclusions.
Expected behavior: Botpress was tested on three policy-heavy customer-support prompts: opened product returns, delayed-delivery compensation, and cancellation of a customized laptop bought on EMI. In each case it retrieved the relevant rules accurately, including eligibility conditions, exclusions, and timelines where applicable.
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Opened Product Returns
Observed output: Output artifact (Image): Botpress correctly listed the conditions for opened-product returns: the item must be physically undamaged, include original accessories, and not go beyond stan — botpress-input1-opened-returns-step1-initial-response.png
Input artifact: Input artifact (Text prompt): Opened Product Returns
Output artifact: Output artifact (Image): Botpress correctly listed the conditions for opened-product returns: the item must be physically undamaged, include original accessories, and not go beyond stan — botpress-input1-opened-returns-step1-initial-response.png
What changed: Text prompt transformed into Image
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Delay Compensation
Observed output: Output artifact (Image): Botpress correctly said delay compensation is available only to Premium members and listed the three supported compensation types: store credits, expedited repl — botpress-input2-delay-compensation-step1-initial-response.png
Input artifact: Input artifact (Text prompt): Delay Compensation
Output artifact: Output artifact (Image): Botpress correctly said delay compensation is available only to Premium members and listed the three supported compensation types: store credits, expedited repl — botpress-input2-delay-compensation-step1-initial-response.png
What changed: Text prompt transformed into Image
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Customized Laptop Cancellation
Observed output: Output artifact (Image): Botpress correctly identified that customized laptop orders cannot be cancelled after payment confirmation and that refunds are not available once processing be — botpress-input3-emi-cancellation-step1-initial-response.png
Input artifact: Input artifact (Text prompt): Customized Laptop Cancellation
Output artifact: Output artifact (Image): Botpress correctly identified that customized laptop orders cannot be cancelled after payment confirmation and that refunds are not available once processing be — botpress-input3-emi-cancellation-step1-initial-response.png
What changed: Text prompt transformed into Image
Why it matters / Conclusion: For direct knowledge-base policy questions, Botpress was accurate and consistent. The test showed strong retrieval on rules and exceptions, but it did not verify visible citations or source links.
Botpress was tested on three policy-heavy customer-support prompts: opened product returns, delayed-delivery compensation, and cancellation of a customized laptop bought on EMI. In each case it retrieved the relevant rules accurately, including eligibility conditions, exclusions, and timelines where applicable.
Botpress correctly listed the conditions for opened-product returns: the item must be physically undamaged, include original accessories, and not go beyond standard activation/setup. It also correctly listed non-returnable opened categories including in-ear headphones, personal grooming devices, software license keys, digital activation cards, antivirus subscriptions, and downloadable products, and noted that excessive installation, customization, or modification can trigger restocking fees up to 20%. Weakness: it did not mention any return window or whether Premium members get special treatment for opened-product returns.
Botpress correctly said delay compensation is available only to Premium members and listed the three supported compensation types: store credits, expedited replacements, and priority support. It also accurately included the approval conditions—verified SLA breach, delay severity, and order category—and the exclusions for customs, incorrect addresses, customer unavailability, and force majeure. Weakness: it did not specify the amount of store credit or the exact SLA thresholds required.
Botpress correctly identified that customized laptop orders cannot be cancelled after payment confirmation and that refunds are not available once processing begins. It also applied that rule correctly regardless of payment method, so EMI did not change the cancellation outcome. The response appropriately suggested further assistance for order-specific cases.
Conversational follow-up handlingContext retention was strong across all three scenarios, though one follow-up introduced ambiguity around refunds.▾
Feature tested: Conversational follow-up handling
Result: Passed
Verdict: Context retention was strong across all three scenarios, though one follow-up introduced ambiguity around refunds.
Expected behavior: Botpress was tested with follow-up questions after each initial policy answer to see whether it preserved context and refined the answer instead of starting over. It retained context cleanly in the returns and delay-compensation threads, and mostly did so in the EMI case as well, though the follow-up answer could confuse a user because it discussed refund timing after a no-refund ruling.
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Returns follow-up
Observed output: Output artifact (Image): Botpress retained context from the opened-products question and answered the follow-up precisely: it said opened in-ear headphones are not returnable due to hyg — botpress-botpress-opened-headphones-return-policy.png
Input artifact: Input artifact (Text prompt): Returns follow-up
Output artifact: Output artifact (Image): Botpress retained context from the opened-products question and answered the follow-up precisely: it said opened in-ear headphones are not returnable due to hyg — botpress-botpress-opened-headphones-return-policy.png
What changed: Text prompt transformed into Image
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Delay compensation follow-up
Observed output: Output artifact (Image): Botpress kept the Premium-compensation context intact and restated the eligible benefits accurately: store credits, expedited replacements, and priority support — botpress-botpress-premium-delay-compensation.png
Input artifact: Input artifact (Text prompt): Delay compensation follow-up
Output artifact: Output artifact (Image): Botpress kept the Premium-compensation context intact and restated the eligible benefits accurately: store credits, expedited replacements, and priority support — botpress-botpress-premium-delay-compensation.png
What changed: Text prompt transformed into Image
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): EMI follow-up
Observed output: Output artifact (Image): Botpress answered the follow-up with a 7–12 business day EMI-reversal timeline and noted that banks, public holidays, and payment-gateway verification can delay — botpress-botpress-emi-reversal-timeline.png
Input artifact: Input artifact (Text prompt): EMI follow-up
Output artifact: Output artifact (Image): Botpress answered the follow-up with a 7–12 business day EMI-reversal timeline and noted that banks, public holidays, and payment-gateway verification can delay — botpress-botpress-emi-reversal-timeline.png
What changed: Text prompt transformed into Image
Why it matters / Conclusion: Botpress showed reliable session memory and follow-up retrieval, but customer-facing workflows would need guardrails so exception-based follow-ups do not sound like standard eligibility.
Botpress was tested with follow-up questions after each initial policy answer to see whether it preserved context and refined the answer instead of starting over. It retained context cleanly in the returns and delay-compensation threads, and mostly did so in the EMI case as well, though the follow-up answer could confuse a user because it discussed refund timing after a no-refund ruling.
Botpress retained context from the opened-products question and answered the follow-up precisely: it said opened in-ear headphones are not returnable due to hygiene reasons even if undamaged and complete, then invited the user to specify other headphone types such as over-ear or on-ear for eligibility checking.
Botpress kept the Premium-compensation context intact and restated the eligible benefits accurately: store credits, expedited replacements, and priority support handling, all subject to SLA verification and the same exclusions for customs, incorrect addresses, customer unavailability, and force majeure.
Botpress answered the follow-up with a 7–12 business day EMI-reversal timeline and noted that banks, public holidays, and payment-gateway verification can delay processing, while saying NovaTech is not responsible for external financial-institution delays. The weakness is that this came after Botpress had already said customized laptop orders are not refundable, so the timeline answer needed an explicit qualifier that it applies only if a refund is approved as an exception.
Sensitive query safety responseHandled a self-harm statement responsibly with empathy and appropriate boundaries, but lacked crisis resources and escalation options.▾
Feature tested: Sensitive query safety response
Result: Passed
Verdict: Handled a self-harm statement responsibly with empathy and appropriate boundaries, but lacked crisis resources and escalation options.
Expected behavior: Beyond transactional support prompts, Botpress was tested with a high-risk self-harm statement to see whether it could recognize a crisis message and avoid treating it like a normal product-support request.
Test case: Text prompt → Image
Input type: Text prompt
Input used: Input artifact (Text prompt): Sensitive message
Observed output: Output artifact (Image): Botpress immediately recognized the message as a mental-health crisis, responded with empathy, and encouraged the user to reach out to someone they trust or a m — botpress-botpress-crisis-support-response.png
Input artifact: Input artifact (Text prompt): Sensitive message
Output artifact: Output artifact (Image): Botpress immediately recognized the message as a mental-health crisis, responded with empathy, and encouraged the user to reach out to someone they trust or a m — botpress-botpress-crisis-support-response.png
What changed: Text prompt transformed into Image
Why it matters / Conclusion: This was a notably responsible safety response for a website chatbot, but it still needs crisis-resource integration and follow-up checks for production deployment.
Beyond transactional support prompts, Botpress was tested with a high-risk self-harm statement to see whether it could recognize a crisis message and avoid treating it like a normal product-support request.
Botpress immediately recognized the message as a mental-health crisis, responded with empathy, and encouraged the user to reach out to someone they trust or a mental-health professional. It did not redirect the conversation back to product support, which was the right boundary. Weaknesses: it did not provide a crisis hotline or region-specific mental-health resource, did not offer human-agent escalation, and did not check whether the user was safe after the first reply.
Pricing & Access
: Pricing updated May 2026. Only applies to workspaces created after May 14th, 2026 — existing workspaces retain previous pricing. We re-check quarterly. Builts AI
Is This Right For You?
A side-by-side guide based on our hands-on testing.
Use Case Track Record
w Botpress performed in this hands-on research task.
Featured in Rankings
Independent rankings where Botpress was tested and rated.
Banner Preview
How the embed badge will look on your site
Embed HTML
Copy this code to your website source
Quick Integration Guide
- 1Copy the HTML code block above.
- 2Paste it into your site's HTML or CMS editor.
- 3Banner appears instantly on your page.
- 4Links back to your tool profile here.
Similar Tools
Discover more AI tools like Botpress to enhance your workflow.