New request
Explored 12 files, 4 searches
I generated the page, connected the data source and added a table sorted by year. The render is ready on the right.
Navin Research
We build agents that write, test and ship software with you.
| 2026 | Semantic code indexing | Published |
| 2026 | Tool-augmented search | Published |
| 2025 | Online reinforcement learning | Published |
| 2024 | Multiple agents in parallel | Published |
~/navin $ navin agent
Session open - 3 tools loaded
Repository scan: 128 files indexed
Ready. Describe your task.
Accelerate development by handing tasks off to Navin, while you keep control over the decisions that matter.
Learn more about agentic developmentRework the sign-up funnel
Navin reads the repository, proposes a three-step plan and waits for your approval before writing any code.
Tasks 3
Extract the form into a component
Add server-side validation
Write the flow tests
Each agent gets its own environment to build, test and ship a feature end to end, without blocking your machine.
Learn more about agentsThis week
Total time worked: 14m 22s
Terminal, browser, team chat: Navin follows you wherever the work happens and keeps the same context everywhere.
Read the documentation$ navin agent
Agent ready. Type your request.
Schedule agents that run continuously to maintain, fix and evolve your software, even when you are away.
Learn more about AutomationsNightly dependency review
ActiveTriggers
Every day at 03:00 - and on every new pull request
Instructions
Check for security updates, run the test suite and open a pull request if everything passes.
360° Vision
Navin's workbench brings code, live preview, project graph, tasks, evolutions and the 360° audit into a single window. Everything moves in real time while the agent works.
Modules
Documents, marketing, SEO, leads, team and infrastructure: every module opens with a slash command and delivers a finished result.
Generate polished PowerPoint, Word, PDF and Excel documents from templates: pitch decks, business plans, contracts, invoices, resumes, sales proposals.
Campaigns, ad videos, product images and social content, produced end to end.
Audits, keyword research, optimized content and link strategy.
Find companies, decision makers and jobs; qualify, enrich, and turn them into pipeline.
Design the organization, staff it, run it - and spawn virtual AI teams on missions.
DevOps, SysOps and infra end to end: Kubernetes, CI/CD, GitOps, cloud, servers, incidents.
The full development workbench - code, preview, graph, tasks and the 360° audit - gets its own section above.
Creative studio
Navin produces brand visuals, ad creatives and complete videos with the best engines on the market, without leaving your conversation.
Product images, brand visuals and social creatives via gpt-image, DALL-E, Imagen and Gemini Flash - with your key, on the provider of your choice.
Scripts, storyboards, then final videos with Google Veo and Sora, in 16:9 for the web or 9:16 for social media.
product-launch-spot.mp4 - generated by Navin
And voice: built-in audio transcription (AssemblyAI) to dictate requests or transcribe your meetings.
Agent teams
Navin ships with a roster of ready-to-work specialists. Mention them in chat, or spawn a full virtual team on a mission.
Team org chart
Fully editable org chart: rename, add or specialize your members.
Virtual team on a mission
RunningLaunch a SaaS product: market research, landing page, email sequence and launch plan.
Claw Agents
160+ built-in skills, combinable into agent packs: prospecting, scraping, investment decks, finance and much more.
Lead generation, qualification, buying signals and enrichment, down to a CRM-ready CSV.
Web extraction, deep research, automated browsing and tender monitoring.
Investor pitch deck, business plan, 3-year financial model and due diligence pack.
Invoice reading, contract extraction, Stripe operations and pipeline analysis.
Install new skills with one command, publish your own and share them with the community on clawhub.ai.
$ npx clawhub install lead-generationA sample of the 160+ built-in skills - each one activates on demand, with zero configuration.
Models & providers
GPT, Claude, Gemini, Mistral, DeepSeek... and your local models. Navin routes every task to the right model and fails over automatically.
30 providers, from cloud to local
Automatic routing per task
Local models: Ollama, LM Studio, vLLM
Automatic fallback on outages
“Navin replaced three internal tools. My agents run on our servers, with our keys, and I see exactly what they do.”
Sarah Lemaire
Platform engineering lead
“Owning the agent changes everything. We adapt it to our repo and our conventions without waiting on a vendor roadmap.”
Karim Benali
CTO, product studio
“Nightly automations handle the boring maintenance. We show up in the morning to pull requests ready for review.”
Julia Ferreira
Senior engineer
“Setup takes a minute and the agent already knows our codebase. It is the first tool my team adopted with zero training.”
Thomas Nguyen
Head of development
“We pay our own inference costs and keep data in house. In a regulated context, that was the entry requirement.”
Amina Haddad
CTO, healthcare
“Switching models per task cut our bill in half without losing quality.”
David Okonkwo
AI infrastructure engineer
Route each request to the right model: long reasoning, fast code, or a local model for sensitive data.
Run several agents in parallel, each in its own environment, and follow their progress in real time.
A minimal, readable core that you can audit, extend and host yourself, for as long as you need.
June 2026 · Engineering
How we run multiple agents without melting your machine
8 min read
May 2026 · Research
Semantic indexing: giving the agent a memory of the repo
11 min read
April 2026 · Product
Automations: delegating the maintenance nobody enjoys
6 min read
March 2026 · Security
Why your API keys never leave your machine
7 min read