How We Got Here
We started building NeuroGen in late 2025 with two tools: a file converter and a web scraper. The idea was simple. People needed to convert documents between formats and pull data from websites. We figured we could wrap some Python libraries in a Flask app and call it a day.
That lasted about three weeks. Once we added the AI assistant, everything changed. Users wanted their assistants to send emails. They wanted them to search databases. They wanted them to remember previous conversations and learn from uploaded documents. Every feature request pointed in the same direction: people did not want a single tool, they wanted a platform.
So we built one. As of today, NeuroGen has 18 production tools, 100+ API integrations, and a multi-tenant system that lets you white-label the whole thing under your own brand. The full picture looks like this.
The 18 Tools
Each tool is its own section of the platform with dedicated UI, APIs, services, and database tables. They are not bolted-on features. They are complete systems.
Some of these tools are tightly connected. The AI Assistant feeds into Chat Integrations, which lets you deploy assistants to Telegram, Discord, Slack, and 12 other platforms. The Knowledge Base powers every assistant and agent with document-backed answers. The Credit system tracks usage across all of them with per-model, per-token precision.
100+ API Integrations via the Marketplace
API Marketplace
Both the Assistant Builder and Agent Builder include a built-in API Marketplace at Step 4 of their wizard. We ship 100+ OpenAPI specs that get parsed at runtime. Your agents can call SendGrid, Stripe, Google Maps, Mailchimp, GitHub, Jira, Notion, Slack, and dozens more. Each agent can connect to up to 10 integrations with 20 tools per integration. Auth is handled per-integration: bearer tokens, API keys, query params, or full OAuth2 flows. Every key is Fernet-encrypted at rest. Every outbound call is SSRF-protected.
The integrations span 19 categories: communication, productivity, developer tools, finance, Google services, CRM, data, logistics, lead generation, maps, weather, ecommerce, business, content, healthcare, legal, real estate, education, and hospitality. We are adding more every week based on user requests.
Credit System and Pricing
Everything in NeuroGen runs on credits. One credit equals one cent ($0.01). When your assistant sends a message, we calculate the exact token count, apply the model's per-million pricing with our 20% markup, and deduct the right number of credits. Same for image generation, TTS, voice calls, and API actions.
We spent a lot of time getting this right. The credit calculator knows the pricing for every model we support (OpenAI, Anthropic, Google, zai, and more). It tracks input tokens, output tokens, cached tokens, audio tokens, and image dimensions separately. No flat rates, no guessing.
Check out the full breakdown on our pricing page.
Enterprise Multi-Tenant: White-Label the Whole Platform
This is the big one. If you run an agency, a consultancy, or a SaaS company, you can now deploy NeuroGen under your own brand. Set your logo, your favicon, your color palette, your custom CSS, and even your own domain. Your customers log into a portal that looks like your product.
Multi-Tenant Architecture
Organizations have a full role hierarchy: owner, admin, manager, member, and viewer. There are 12 granular permissions you can assign per role. Credit delegation works on three tiers: org pools feed into customer allocations, which then deduct at the individual level. You get analytics dashboards, CRM integration with Mautic and SuiteCRM, admin impersonation for support, and a migration tool to move existing solo users into your organization.
We built this in 12 phases over the course of February. The database foundation went in first, followed by the RBAC system, customer portals, credit delegation, white-label config, branded standalone chat, analytics, CRM hooks, impersonation, and finally the migration tooling. It is production-ready today on the Enterprise tier.
What Changed Under the Hood
Beyond the visible features, we did a lot of infrastructure work this month. NeuroGen Cortex — our knowledge layer — now does adaptive multi-step retrieval and automatic query decomposition for complex questions. It powers every knowledge base interaction on the platform.
On the credit system side, we corrected 251 usage log entries, fixed float-to-integer rounding issues, added with_for_update() locks on balance changes, and validated that all margins match our 20% markup target. A daily reconciliation job at 2 AM UTC pulls actual costs from OpenAI and Anthropic billing APIs and compares them against our internal logs. If the discrepancy exceeds thresholds, the admin gets an alert.
We also hardened the production stack: every JSON column has a _safe_json() wrapper, every commit is wrapped with rollback, and error logs no longer leak exception details to users. The security audit added global CSRF protection, webhook signature verification, IDOR fixes across 16 API endpoints, HMAC-signed share cookies, and SSRF protection on all outbound HTTP calls.
What We Are Building Next
We are not slowing down. The roadmap for March includes OAuth2 expansion for QuickBooks and Shopify integrations, a visual workflow builder for multi-step agent pipelines, and a public API so you can build on top of NeuroGen programmatically.
We are also working on a self-hosted deployment option for teams that need to keep everything on their own infrastructure. More on that soon.
If you have been watching from the sidelines, now is a good time to jump in. The demo tier is free with 100 credits. That is enough to build an assistant, connect a knowledge base, and test a few API integrations. No credit card required.
Try NeuroGen Today
Start with a free demo account. 100 credits, all 18 tools, no time limit. Build something and see what the platform can do.
Start Free Trial