Browse categories
Explore
Fiverr Pro
English
$
USD
I will be programming ai receptionist by ai agent developer
Stas Sorokin
Level 2
About this gig
Turn tasks into autonomous workflows.
I build agent-driven systems using AgentKit (or equivalent frameworks) combined with the power of n8n with curated tools (search, database, APIs), memory modules, safe action-gating, and orchestration logic built in.
What you'll get:
- An agent that can reason, act, and learn: connect tools like web search, DB reads/writes and API calls into one smart workflow
- The agent retains state/data across steps so its not just a trigger action-bot, but one that tracks progress and revisits context
- Safeguards on tool access, human approval nodes, retry/fallback logic, so your agents perform reliably and safely in production
- Workflows that coordinate multiple agents, schedule tasks, handle data pipelines, monitor status, log activity, and integrate with your systems
Why this matters:
Traditional automations are simple triggers actions. But agentic workflows use memory, reasoning, tool-use and orchestration to think, not just do. Modern platforms like n8n support AI agents that can integrate with your stack at scale.
Ready to automate like you've never before? Let's build your next-gen autonomous workflow secure and extendable.
Get to know Stas Sorokin
Stas Sorokin
Expert Software Development Full Stack Lead Engineer
Level 2
- FromIsrael
- Member sinceJul 2017
- Avg. response time1 hour
- Last delivery2 days
Languages
English, Russian
Welcome!
I'm Stas Sorokin with over 9 years of experience in the tech industry, including work with leading organizations. Specializing in front-end, back-end, native, and mobile development, I've honed a diverse skill set to meet your unique needs.
My record of 50+ positive reviews underscores commitment, quality, speed, and clean code. I believe in putting the client at the center of everything I do. From our initial consultation to the final delivery, you will get a high-standard service that exceeds your expectations.
Let's collaborate together to turn your ideas into a digital reality!
Among my clients
Vercel
Software Development
Fiverr
Internet Software & Services
My Portfolio
Other Software Development Services I Offer
FAQ
Which tools and systems can you integrate with?
I can connect databases, CRMs, email, calendars, APIs, and internal systems. Typical tools include Google Workspace, HubSpot, Slack, Airtable, Notion, and any custom HTTP or REST endpoint.
How do you handle AI hallucinations or errors?
Agents are configured with logic guardrails, validation layers, and model evaluation tests. This ensures the AI stays factual, with safe fallback responses when confidence is low.
Do the agents have memory or context retention?
Yes. I implement either vector-based long-term memory or structured database memory. This lets agents recall context, past tasks, and user details across conversations or workflows.
Can I monitor agent performance and logs?
Yes. Observability includes traces, metrics, and structured logs. You’ll be able to see when tasks start, succeed, or fail and get insights into latency, token usage, and agent behavior.
Are AI operation costs optimized?
Yes. Agents are tuned for minimal token usage and caching efficiency. I use batching, truncation, and prompt optimization to reduce API costs while maintaining quality and reliability.
Can the agent work offline or without internet?
Limited offline capability is possible. Tasks are queued and processed when connectivity is restored. Agents requiring real-time model calls need an active internet connection.
How is authentication handled for integrations?
OAuth 2.0, API keys, or JWT-based auth are supported. Credentials are stored securely in environment variables or encrypted vaults following least-privilege access principles.
How do you ensure system security?
Security follows least-privilege design. All integrations use HTTPS, scoped tokens, and sanitized logs. Sensitive data is never exposed or stored beyond what’s operationally necessary.
Can humans intervene during agent execution?
Yes. I can enable a human-in-the-loop setup, where certain decision points trigger manual approval or review, giving you oversight before critical actions are executed.
Can the system scale for heavy workloads?
Yes. The architecture supports horizontal scaling using worker queues and async execution. It can handle parallel agents or workflows across multiple instances efficiently.
