startupregistry
Record SR-00019·First registered 24 Aug 2026

custodianlabs.io

Custodian Labs | Build and deploy AI agents in 5 lines of code

Recorded wording

102 lines, exactly as the page had them.

Snapshot24 Aug 2026, 06:15 UTC
  1. 01Custodian Labs | Build and deploy AI agents in 5 lines of code
  2. 02Iframe is blocked. Accept cookies to load it.
  3. 03Manage Consent Preferences
  4. 04Necessary for the site to function. Always On.
  5. 05No trackers detected for this category.
  6. 06No providers detected for this category.
  7. 07Measures usage and improves your experience.
  8. 08Used for targeted advertising.
  9. 09Remembers your preferences and provides enhanced features.
  10. 10Thank you! Your submission has been received!
  11. 11Oops! Something went wrong while submitting the form.
  12. 12By clicking "Accept" , you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy
  13. 13for more information.
  14. 14Pre-seed funded · AUT Ventures & New Zealand Government
  15. 15Build and deploy AI agents in 5 lines of code.
  16. 16Custodian Labs takes you from idea to deployed, production-grade AI agent — with privacy built in at every layer. No infrastructure. No data leakage.
  17. 17from custodian_labs import Custodian
  18. 18system_prompt="You are a helpful assistant...",
  19. 19Everything you need. Nothing you don't.
  20. 20The proprietary Guardian Layer is the guardrail that detects PII and puts you in control of how it's handled before it reaches any model.
  21. 21Give your agent long-term memory and document retrieval without configuring embeddings or vector DBs. Add a knowledge base in one line and it just works.
  22. 22Switch between OpenAI, Anthropic, Mistral, or local models with one line — no vendor lock-in, no rewrite required. Your agent logic stays exactly as it is; only the model underneath changes.
  23. 23Get started for free
  24. 24No database to provision, no vector store to connect, no hosting to manage. Custodian abstracts every layer beneath your agent so you focus entirely on logic.
  25. 25Research-grade foundations
  26. 26Commercialised with AUT Ventures and New Zealand Government funding. Production reliability grounded in deep academic research.
  27. 27The Custodian Difference
  28. 28No databse. No hosting. Just deploy.
  29. 29Normally you'd provision a vector DB, set up a hosting environment, wire up your model provider, write retry logic, handle streaming — all before writing a single line of agent logic. Custodian collapses that entire stack to a function call.
  30. 30# Step 1: Provision a vector DB + embedding pipeline
  31. 31pinecone.init(api_key=...), index = pinecone.Index("my-index")
  32. 32embedder = SentenceTransformer("all-MiniLM-L6-v2")
  33. 33chunks = [text[i:i+512] for i in range(0, len(text), 448)]
  34. 34index.upsert([(f"doc_{i}", embedder.encode(c)) for i, c in enumerate(chunks)])
  35. 35# Step 2: Manual PII scrubbing (fragile regex, hope for the best)
  36. 36re.sub(r"[A-Z][a-z]+ [A-Z][a-z]+", "[REDACTED]", prompt)
  37. 37# Step 3: Tool dispatch + retry logic
  38. 38def call_with_retry(messages, max_retries=3):
  39. 39for attempt in range(max_retries):
  40. 40try: return client.chat.completions.create(...)
  41. 41except: time.sleep(2 ** attempt)
  42. 42# Step 4: Session memory via Redis
  43. 43redis_client.setex(f"ctx:{session_id}", 3600, json.dumps(messages[-20:]))
  44. 44# Step 5: Flask server, auth, rate limiting, CORS, logging...
  45. 45# Step 6: Write your Dockerfile, k8s config, CI/CD pipeline...
  46. 46# Step 7: Finally — your agent runs. Maybe.
  47. 47Multi-agent capabilities
  48. 48Assemble your team. Deploy as one.
  49. 49Build teams of specialised agents that each handle their own domain — then deploy them as a single unified endpoint. Custodian's router automatically directs each query to the right agent.
  50. 50Give each agent its own data sources, tools, system prompt, and topic scope. Scale the team without rewriting the interface.
  51. 51Technology & engineering questions
  52. 52Intelligence & strategy queries
  53. 53Science & research topics
  54. 54Automatically routes every query to the right agent
  55. 55from custodian_labs import Agent, AgentTeam
  56. 56name="the_engineer",
  57. 57system_prompt="You are a senior engineer. Handle all tech and engineering questions.",
  58. 58topics=["technology", "engineering", "AI"],
  59. 59name="the_strategist",
  60. 60system_prompt="You are a strategy lead. Handle intelligence and strategy questions.",

Showing the first 60 of 102 recorded lines · Lines under 20 characters are not recorded: too short to prove anything

Register entry

ReferenceSR-00019
Entryunclaimed
First registered24 Aug 2026
Snapshot24 Aug 2026
Domain verifiednot yet
Watchnone
Last sweep24 Aug 2026
Matches found0

Sweep

Find out who else is using your wording.

We search the open web for your exact lines, read every page we find, and show you what overlaps. It takes a minute or two and you watch it happen.

The first sweep on a record is free.

See the whole register

Sweep history

24 Aug 2026, 06:16 UTC15 pages read0 matchesOpen
24 Aug 2026, 06:15 UTC15 pages read1 matchesOpen

Unclaimed entry

This record was built from the public page, not by its founder.

Nothing here is invented: the wording is what the page said on the date shown, and there are no numbers we were not given. If it is your domain you can take it over, and if you would rather it were not here at all, say so and it comes down within 24 hours.

Removal requested 24 Aug 2026, 06:20 UTC. This record comes down within 24 hours.

custodianlabs.io · SR-00019 — startupregistry