Maria Siasat.

Maria Siasat

Current

B2B Growth Strategists

b2bgrowthstrategists.com

Projects

Experimenting with AI agent teams to scale businesses

The goal: have fun with every AI tool I can get my hands on, start a business in 24hrs, and have my agent team run it. One rule: no outside technical help, unless I am reselling a platform. This is the running log of the experiments, the tools behind them, and the mistakes I made.

The marketing agent creator club

Marketers tinkering with agents and marketing workflow to grow real businesses.

The setup

One person, six agents, one workflow.

Research agentSEO / AEODrafting agentCONTENTOutbound agentPROSPECTINGVoice agentLIVE COACHINGEmail agentNURTURECFO agentNUMBERSMariaSiasat

Experiment 01

Review Optimiser AI

Done for you Trust Management.

Can an AI-orchestrated agent team grow and scale a business on its own?

Build process

  1. 01Landing page built with a Lovable prompt.
  2. 02Inbound webhooks created in Zapier.
  3. 03Outbound prospecting and demand gen via an Instantly email nurture sequence, with the goal to close.
  4. 04SEO and AEO agent built on Claude, optimising search visibility.
  5. 05ICP research and pipeline gen via Helena, a digital marketing agent built in OpenClaw by Enrich Labs.

Integrations used

  • Lovable
  • Lovable Cloud (Postgres, auth, storage)
  • Zapier webhooks
  • Instantly
  • Claude
  • OpenClaw / Helena by Enrich Labs
  • Google Business Profile reviews
  • Resend email
  • Agent 1: SEO and AEO agent
  • Agent 2: Outbound prospecting agent

Status

Built. Demand gen to commence.

Start
20 July
Finish goal
20 August
Delayed
Not yet, 22 days to go
www.reviewoptimiser.com

Finding

The front end took two hours to build. Connecting the outbound agent to Instantly, and testing the sequencing, cost four to five days. That gap, not the creative work, is the real bottleneck for a non-engineer.

Keep me posted

Experiment 02

Bramwell AI

SaaS voice agent communication coach. Next evolution: an all-in-one coaching platform.

Integrations used

  • Lovable
  • Lovable Cloud (Supabase Postgres, auth, storage)
  • Realtime voice AI
  • Speech to text transcription
  • Stripe subscriptions and paid tiers
  • Edge functions for session scoring

Status

Delayed.

Start
6 April
Finish goal
2 July
Delayed
27 days over
www.bramwellai.com

Finding

Same pattern. The coaching logic came together fast. Connecting the APIs, voice, database and payments, and testing them as one system, is what actually slowed things down.

Keep me posted

Experiment 03

Business Trust Council

Independent trust and credibility research, published as reports.

Can an AI agent apply a real research methodology, sourcing, sampling and verification, and produce reports that hold up to scrutiny?

Integrations used

  • Lovable
  • Lovable Cloud (Postgres, auth, storage)
  • Claude research agent
  • Structured source verification
  • Automated report generation
  • Resend email
  • Agent 1: Research and methodology agent
  • Agent 2: Report QA agent

Status

In build. Methodology under test.

www.businesstrustcouncil.com

Finding

Accuracy is a sourcing problem, not a writing problem. The agent drafts a clean report in minutes, then verification of every claim is where the days go.

Keep me posted