David MartinAI knowledge systems & contentGet in touch
AI systems

AI systems in daily use.

I designed and built these systems to support content, proposal and sales work at a 40-year Canadian IT firm. Each one is in regular use, and a person reviews every output before it is used.

These are internal company systems, so they are described here rather than shown.

Problems I solve

Problems these systems address.

Four common bottlenecks in content, proposal and sales work, and the system built for each.

The problemEach new proposal takes days to set up.What I do

A structured AI workspace that drafts from approved content and routes gaps to subject experts.

The problemAI pilots do not reach daily use.What I do

A curated knowledge base, defined limits and human review built in from the start.

The problemContent approvals run through email and are hard to track.What I do

One workflow with defined statuses from draft to published, visible to everyone involved.

The problemRelevant bids are missed across multiple procurement portals.What I do

A weekday digest that screens 18 sources and sends relevant opportunities to leadership.

Built with Claude on SharePoint

Proposal Guru

The problem
Each new RFP required 1 to 2 days of setup, followed by weeks of drafting across several authors with different writing styles.
What I built
A structured AI workspace with house style, templates, a tracker and one folder per bid. It flags gaps for subject experts and maintains its own status report.
The result
A new RFP becomes a folder structure, branded templates and an 80 to 90% complete first draft in about 30 minutes.
  1. New RFP
  2. Folders and templates
  3. AI first draft
  4. Expert review
  5. Submission
Built with Claude on SharePoint

Social Media Register

The problem
Post approvals were managed in email threads, with no clear view of what was pending.
What I built
A system that drafts post copy and graphics, then moves each post through review with a checklist and a defined status.
The result
All company social posts now go through it, from draft to publication.
  1. Draft
  2. Awaiting approval
  3. Changes required
  4. Approved
  5. Scheduled
  6. Published
Microsoft Copilot Studio

Company AI agents

Agents for specific functions: bid response, marketing, Google Ads, cybersecurity advice, sales sequences and 3D image prompts.

In a leadership session, the bid agent produced a 35-page RFP response live.

Weekday digest

Bid Radar

A weekday digest of new public-sector opportunities from BC Bid, CivicInfo BC and 16 buyer portals, sent to the 11-person leadership and sales group. Revised four times in its first three weeks based on reader feedback.

I also consolidated 32 years of bid history, more than 1,700 pursuits, into one traceable record.

Supporting tools

Claude projects

Claude projects for one-page sales sheets, video scripts, blog posts, business cards and sector outreach campaigns. Company social posts are written with a Claude project built on a shared knowledge base, and a second project converts each post into a blog article in the house style.

Training

AI training.

I present at the company's AI pilot program and run training on each system I build. The video below is the longer of two training videos on the ELIZA effect.

The ELIZA Effect: How Human Habits Quietly Break AI Collaboration

How human habits quietly break AI collaboration

About five minutes. Built as a presentation and converted to a narrated video with Microsoft 365 Copilot.

  • PowerPoint
  • Microsoft 365 Copilot

Produced for MYRA Systems Corp.

Discuss a role or project.

I design and run AI systems like these for content, proposal and sales teams.

Contact me →