Joined August 2014
- Tweets 255
- Following 185
- Followers 50
- Likes 68
45 Photos and videos
We are investigating unauthorized access to GitHub’s internal repositories. While we currently have no evidence of impact to customer information stored outside of GitHub’s internal repositories (such as our customers’ enterprises, organizations, and repositories), we are closely monitoring our infrastructure for follow-on activity.
1,667
5,301
25,395
13,832,161
May 1
Outsource the work, not the wisdom.
Feb 4
you can outsource your thinking
but you cannot outsource your understanding
27
Apr 22
Claude Code's desktop redesign signals something:
- Traditional dev team scaling
- Parallel agent-powered development
The app architecture picked a side.
Which side are you on?
#AICoding #ParallelAgents #ClaudeCode
37
Apr 21
The real question is which one's your bottleneck. Losing customers faster than gaining them? Features. Strong retention but can't get enough eyeballs? Marketing.
Apr 20
When growing a SaaS,
what’s the priority?
* Features
* Marketing
9
Apr 20
The debugging trap isn't inevitable - teams that establish clear specs before letting AI write the code tend to keep the speed wins without the debugging tax.
Apr 20
Before AI:
- Coding: 3 hours
- Debugging: 1 hour
After AI:
- Coding: 3 minutes
- Debugging: 1 week
9
Apr 20
Q3 AI deployments fail on infrastructure, not intelligence. Rate limits and schema drift kill more production agents than model accuracy ever will. CTOs: monitor APIs, not just benchmarks. #AILeadership #TechStrategy
1
11
Apr 15
OpenAI just killed AI democratization. GPT-5.4 limited to 'select few' users. The best models are becoming exclusive country clubs. Engineering teams betting on open access need new plans. #AIStrategy #TechLeadership
25
Apr 13
'AI kills engineers' misses what's working: hybrid teams. Engineers using Claude for scaffolding human judgment for architecture ship 3x faster. Companies embracing this split outpace those debating replacement. #AILeadership #AgenticWorkflow @claudeai
25
Apr 12
90% fewer vendor meetings for 50 CTOs using a simple AI evaluation framework this quarter. The shift: Stop evaluating features, start evaluating shipping velocity. One question cuts through all the noise: 'Can this ship an agent next week?' #AILeadership #AgenticWorkflow
15
Apr 11
AI dev cycles collapsed: 8 months to 3 weeks. Teams shipping with Cursor Claude Code are 60% faster. Ship with AI, don't debate. #AILeadership #AgenticWorkflow #NativeAI
18
Apr 8
AI coding agents boost teams 4x faster. Stop hiring coders, train spec engineers. Cursor Claude write code, humans design systems. Coding becomes orchestration. #AINative #SpecEngineering
1
42
Apr 7
AI burnout reality check: your devs are coding more, not coding faster. Teams using Claude/Cursor cut overtime 40%, boost delivery 50%. Stop tutorial hell - learn AI on production features. #AIEngineering #AgenticWorkflow
25
Apr 5
We cut feature delivery from 8 weeks to 3 days. AI agents generate 60% of our code - tests, CRUD APIs, docs. Engineers focus on business logic and architecture. Ship with AI, don't debate it. #AIEngineering #TechLeadership
1
1
13
Apr 4
Claude spotted a Linux kernel vulnerability that experts missed for 9 years. Real ROI of AI code review isn't speed - it's catching the critical edge cases humans don't see.
mtlynch.io/claude-code-found…
#AICodeReview #TechLeadership
19
Apr 1
this thing has my debugging sessions figured out. it's also a COMMON rarity which somehow feels personal
go type /buddy in Claude Code right now. your pet won't count toward usage either 👀
47
Mar 31
Your AI agents failed because you skipped architecture. Production needs 4 layers: Context, Tools, Memory, Guardrails. Single-layer chatbots don't scale. This architecture does. #AILeadership #ProductionAI #AIEngineering
12
Mar 30
Your AI rollout failed because you bolted agents onto human processes. I've rebuilt SDLCs for AI-first development - teams get 3x productivity when you design workflows around AI reasoning, not handoffs.
#AIEngineering #AITransformation
20
Mar 30
AI fails when you digitize dysfunction. Don't automate broken processes - fix them first, then scale.
#AILeadership #TechDebt
10
Mar 29
Claude Code doesn't just write code.
It reads your codebase, matches your patterns, runs tests, fixes bugs, and explains every decision.
That's not autocomplete. That's a dev who never sleeps.
#ClaudeCode #AI
1
18
Mar 26
Giving Claude agents tool access (bash, file reads, MCP) changed how I think about dev work.
It's not coding assistance anymore. It's delegation.
You need architect instincts, not typing speed.
#ClaudeCode #AI
1
1
21