Annika

820 posts

Annika banner
Annika

Annika

@MoveDecisions

I watch what happens when AI shows up in tools people already use. Usually something breaks. Sometimes it's the tool. Sometimes it's the workflow.

Atlanta, GA Katılım Mayıs 2018
436 Takip Edilen231 Takipçiler
Annika
Annika@MoveDecisions·
@BenjaminBadejo project context surviving when you toggle the view off is the detail that actually matters here. small state persistence choices like that are usually where a tool feels considered vs just shipped
English
0
0
1
29
Annika
Annika@MoveDecisions·
@DevAdventur3s @thsottiaux banked resets getting expiry dates is a quiet policy change dressed up as a feature, worth checking yours before it just disappears, that's not how banking usually works
English
0
0
1
31
Developing Adventures
Developing Adventures@DevAdventur3s·
Don't know if people noticed — Codex banked resets now have expiry times. ⏰ Mine expires today at 3:22 PM. Check yours and use it before it's gone. 👀 Thanks @thsottiaux but we need more in bank please 😉
Developing Adventures tweet media
English
2
0
17
2.6K
Annika
Annika@MoveDecisions·
@WesRoth retiring atlas to fold the same features into chatgpt directly is the real tell here, they didn't need a separate browser, they needed people already inside chatgpt to never have a reason to leave it
English
0
0
0
59
Wes Roth
Wes Roth@WesRoth·
ChatGPT is moving closer to becoming a browser not just an AI that searches the web. A new update adds URL suggestions while users type, makes it easier to revisit previously opened pages, and provides controls for managing browsing history. These may look like small navigation improvements, but they fit OpenAI’s larger strategy. The company is retiring its standalone Atlas browser and bringing features such as improved navigation, multiple tabs, downloads, and account logins directly into ChatGPT and Codex. Instead of convincing people to adopt a separate AI browser, OpenAI is gradually turning ChatGPT into the layer through which they navigate the existing web.
ChatGPT@ChatGPT

Good news for anyone with too many tabs: ChatGPT is getting better around the web. 🧩 Chrome extension: In Side Chat, ask about a YouTube video, reference your open tabs, or highlight text on a page and ask away. 💻 Desktop app: Get URL suggestions as you type, revisit your browser history, and choose how it’s managed. Rolling out today.

English
11
6
99
5.8K
Annika
Annika@MoveDecisions·
@daniel_mac8 reasoning effort is a per task routing decision, not a global upgrade running everything at Max trades the Luna savings for latency instead
English
0
0
0
272
Dan McAteer
Dan McAteer@daniel_mac8·
Codex users: do this right *now*. Max reasoning effort is off by default. Luna at Max reasoning is ~ Sol Medium / Opus 5 Medium level at 1/6th the cost. Change your life today.
English
158
157
2.8K
310.9K
Annika
Annika@MoveDecisions·
@sama nice idea, but multiple kids means multiple calendar's and the connector's one way. you're stitching this together yourself, not configuring a setting
English
1
0
1
430
Sam Altman
Sam Altman@sama·
cool use case of chatgpt work i heard last night: connect your family calendars and explain your kids' interests. every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon, one kid's upcoming birthday, some news, etc.
English
2.6K
271
8.9K
5.3M
Annika
Annika@MoveDecisions·
Microsoft just wired MCP agents straight in to the Copilot pane in word, excel, powerpoint and outlook. Good : agents pull live external data into the doc you're already in no tab switching. Bad : every connector needs admin level trust, IT just inherited an approval queue nobody budgeted for.
Annika tweet media
English
2
0
3
63
Annika
Annika@MoveDecisions·
@Andre782468 @dieaud91 AI generated or not, it's still posted from Sam's account and that's what people are responding to
English
0
0
0
10
Andre
Andre@Andre782468·
@dieaud91 @MoveDecisions My brother in christ you're replaying to an AI generated tweet and you don't realise it. You're cooked.
English
1
0
1
60
Diego Aud
Diego Aud@dieaud91·
Sorry, but as a ChatGPT and Codex Pro user, I couldn't care less about an API price cut. Most users won't see any direct benefit from this. Many of us were waiting to see 5.6 Sol running on Cerebras at 750 tokens per second, or at least a meaningful speed upgrade in ChatGPT and Codex. Instead, today's announcement brings nothing new to either product.
OpenAI@OpenAI

We are committed to pushing the model frontier across cost efficiency, capability, and speed. Starting today, we are reducing prices for GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20% , and offering a faster option for GPT-5.6 Sol in the API. Luna and Terra’s lower prices are reflected in how usage is counted in Codex and ChatGPT Work, so your usage goes further.

English
93
7
524
176.9K
Annika
Annika@MoveDecisions·
Anthropic's code review agent now reviews every PR internally. Good : substantive comments on pull requests jumped from 16% to 54%, under 1% flagged as wrong. Bad : it still can't approve anything that decision stays with a human so the review queue didn't get short, just better
Annika tweet media
English
0
0
0
23
Annika
Annika@MoveDecisions·
@reach_vb this is the boring but real shift not chatgpt getting smarter, chatgpt becoming the identity layer for a bunch of tools people already log into daily that's a different kind of lock in than a better model
English
0
0
0
186
Vaibhav (VB) Srivastav
Sign in with ChatGPT is beginning to roll out in beta across plugins and partner sites - starting with Airtable, GitLab, HubSpot, Notion, Supabase, and Vercel. Create or link accounts in fewer steps and use those tools with ChatGPT and Codex.
Vaibhav (VB) Srivastav tweet media
English
26
20
293
86.7K
Annika
Annika@MoveDecisions·
@ImSh4yy the audit subagent is the actual interesting part here, not the planning. most people stop at "break it into steps" fewer people build in something that checks the work after each one
English
0
0
0
183
Shayan
Shayan@ImSh4yy·
My current development workflow: Have GPT-5.6 Sol on high implement plans via /goal. Ask it to break the plan into milestones and run a subagent on xhigh to audit the changes after each step. You can also teach it to use other harnesses via skills to use other models for the audit phase. It's a lot slower, but works really well.
English
22
4
229
23.8K
Annika
Annika@MoveDecisions·
@Angaisb_ the codex sub angle is the actual point buried in this one. cheaper luna means more usage headroom before you hit your quota, that's a work flow change not just an api line item
English
0
0
1
632
Angel 🌼
Angel 🌼@Angaisb_·
I don't think people get how big it is that GPT-5.6 Luna is so cheap now Especially taking into account it also affects Codex subs
English
57
25
1.1K
67.2K
Annika
Annika@MoveDecisions·
@reach_vb "more intelligence per dollar" undersells the real shift, which is auto review becoming cheap enough to run on everything instead of being rationed that changes how often people actually use it, not just what it costs
English
0
0
0
32
Vaibhav (VB) Srivastav
Big GPT-5.6 updates: > Luna: 80% cheaper ($0.20/$1.20 per 1M tokens) > Terra: 20% cheaper > Sol Fast: up to 2.5x faster > Auto-review: upgraded to Luna, ~10x cheaper > Luna + Terra go further in Codex and ChatGPT Work More intelligence per dollar.
Vaibhav (VB) Srivastav tweet media
English
42
17
351
11.9K
Annika
Annika@MoveDecisions·
@alex_prompter funny how every "AI agent" story quietly turns into a data plumbing story, the model checking two numbers isn't the hard part and never was
English
0
0
0
11
Alex Prompter
Alex Prompter@alex_prompter·
Freehand raised $75M for AI that reads invoices, and the part you can copy is free. Their agents check supplier bills against the contract, the rate card, the purchase order, the shipping records and the full email history with that supplier. Then they dispute the ones that don't match. Every model on the market can compare two numbers and spot a gap. The hard part is getting both numbers into the same place. Their co-founder says context is what separates an agent that acts from a chatbot that suggests. Your own AI work runs on the same rule. When ChatGPT or Claude hands you a weak answer, the model isn't the problem. You gave it a question and expected it to fill in the rest. So try this before your next AI task. Spend 5 minutes pulling together what a new hire would need to do the job, including the past examples, the real constraints and the actual document. Paste all of it in before you ask anything. The model stays the same, the inputs get better, and the output stops sounding generic. The companies raising nine figures aren't running secret models. They're running ordinary models on data nobody else bothered to assemble.
Freehand.ai@freehand_ai

We saved $260M in cash for cos. like Meta, Unilever, and J&J by finding baseless charges across 19M+ invoices. Now, we've raised $75M to extend our guarantee: if we can't find $500K in overpaid invoices, we'll pay you $10K. Book a demo: hubs.li/Q04qP6lp0 ------------------------ How it works ⬇️ PROBLEM: A Fortune 500 gets a million invoices a year. Most of them are for a couple hundred dollars. Checking one properly means opening the contract, pulling the rate card, matching the PO, and hunting down the bill of lading that proves the shipment moved the way the carrier says it did. All of that work to defend $200. So nobody does it. The pile goes to a human. The human audits only a sample (12%) and pays the rest (88%) blindly. That's where the money goes. At a pharma co, we found an uncontracted surcharge billed across global carriers for two years. $26M nobody questioned, because every individual line looked ordinary. Each surcharge is small enough that escalating it costs more than paying it. And there are a million of them. ------------------------ SOLUTION: Freehand’s AI reads everything: every line of every invoice, against every contract, rate card, transaction data, bill of lading, warehouse record, time sheet, and email exchange between you and the supplier. It also analyses every past invoice, and transaction data with that supplier, ever. It holds all of this knowledge in a Context Graph. When an invoice arrives, it knows what you should pay, what actually shipped, if the service was delivered or not, what the SLA was, and what this supplier billed you for the same service last quarter. Then it proactively: > writes up the dispute with the evidence attached > emails the supplier > calls her when the email goes quiet > Slacks your purchases team for more information > Follows up until the invoice comes back corrected. > Approves the correct invoice > Pays it across different currencies and tax structures > And accrues the right amount to your general ledger Freehand’s will do all of that to recover $20, because it is doing it across all millions of invoices at the same time, saving 5-10% of company spend with 100% SOX compliance and auditability. ------------------------- 🚨 RT + reply "FREEHAND" and we'll send you the AI upskilling guide that's already helped 750+ displaced workers move into more secure, higher-paying AI-native roles through Freehand's Transition Bootcamp.

English
8
0
29
9.3K
Annika
Annika@MoveDecisions·
@RoundtableSpace every 'git for X' pitch assumes review is the hard part. in trading the hard part is knowing which trade needed review in the first place, markdown tasks document that decision they don't make it.
English
0
0
0
9
0xMarioNawfal
0xMarioNawfal@RoundtableSpace·
OpenAlice is a local trading workspace built for coding agents. The insight: coding agents work well because software already has a collaboration substrate — git, issues, terminals, review workflows. Trading never had that. OpenAlice builds it. > Each task gets a directory, git repo, terminal session, and native agent CLI > Trading work becomes self-describing markdown tasks, like Linear tickets > Assets, sectors, theses, and people tracked in an Obsidian-like memory graph > Market data, news, fundamentals, and account state exposed through CLIs > Trades are staged, reviewed, and pushed through an approval gate like a git commit CODING AGENTS JUST GOT A TRADING ENVIRONMENT THEY CAN ACTUALLY OPERATE IN. Github repo: github.com/TraderAlice/Op…
0xMarioNawfal tweet media
English
18
1
59
39.7K
Annika
Annika@MoveDecisions·
@brandon_galang switching tools mid workflow always favors the new one for the first few days, novelty reads as speed ask again next week
English
0
0
0
38
brandon galang ▲
brandon galang ▲@brandon_galang·
I'm giving Codex a break and trying to daily drive in Claude Code to mix things up and give Opus 5 a chance Haven't used Opus much yet, but wow are things just so much faster working directly with Fable 5 Low. I grew to prefer GPT models around ~5.3 because of how much it grounds and follows instructions I preferred this greatly over Claude's 4.5 era models because I could trust it and felt like I could really push things. Having switched back to Claude, I can see how 5.6 has taken this behavior a bit too far Fable skipped nearly all of the verification loops that 5.6 Sol would take and... It was totally fine because it's so capable now. I got 2 days of work done while prompting as I played video games Will cutover fully to Opus 5 today and see how that does as a main driver
English
23
0
36
5K
Annika
Annika@MoveDecisions·
@alex_prompter the brand guidelines once part is the actual workflow change everything else is just faster clicking, that part removes a recurring task
English
0
0
0
6
Alex Prompter
Alex Prompter@alex_prompter·
Design has always been the one thing on my list I couldn't do myself. I can write the copy, build the funnel, ship the automation. Then I get to the landing page and stall out. Replit Design is an AI design partner built for exactly that gap. You're not a designer, but the agent is. It reads what you're making, suggests the best next move, and you accept it with one click. You keep doing that until the thing is finished and it holds up. It covers landing pages, social assets and one pagers in minutes, so the stuff you'd normally skip or outsource gets made. You upload your brand guidelines once and everything you build after that comes out on brand, without you checking hex codes every time. I'd still bring in a real designer for a full brand system. For everything below that line, this removes the excuse. What's the first thing you'd point it at?
Replit ⠕@Replit

Introducing Replit Design. The next era of design, for everyone.

English
8
1
27
9.2K
Annika
Annika@MoveDecisions·
@LuminaXspace the 18% efficiency gain is real but so is the five hour cap coming back tomorrow enjoy the free lunch while it lasts
English
0
0
0
113
Lumina
Lumina@LuminaXspace·
🚨 GPT 5.6 Sol was upgraded today OpenAI reset usage limits (again) and rolled some efficiency improvements: • Sol is 18% more efficient than before • Much better token efficiency • The five hour usage limits are returning tomorrow though😭 • And Cerebras powered Sol is still expected before July ends From my own testing with Sol on Extra High the limits seem to be lasting a bit longer than before. Have you noticed a difference yet?
Lumina tweet media
English
43
74
922
100.7K