AutogenAI Pricing Overview

AutogenAI are great at introducing you to different parts of their team at the right time. They involve you in future plans and give you opportunities to collaborate to make the tool work better for you as a client. Every person we’ve met in the team has been fantastic and knowledgeable in their area of expertise. Review collected by and hosted on G2.com.

There’s an opportunity to provide more practical advice for the testing and the implementation support.

For example we developed our own spreadsheet for testing the tool. This spreadsheet detailed the content we were testing - how many hours it took to create the content, sources used to create the content, and estimated time saved. It was important for us to validate the tools claims that it could draft 80% faster. In fact, we found in testing it drafted 85% faster than a human could to the same or better quality.

In a similar way, we developed an internal spreadsheet to categorise use cases. This guidance could have been given as part of the implementation with example templates. Review collected by and hosted on G2.com.

I appreciate that the product is regularly updated in response to client feedback. Even though some of the current features might not be exactly what we need at the moment, it's reassuring to know that our input is taken into account and may influence future updates. I think the best three features of the product are:

1. Prompts/Pure Completion: The range/flexibility here has been great in terms of reviewing first drafts for bids, as it provides a more accurate viewpoint of whether questions/specifications have been addressed in comparison to Gamma Review. I was able to design my own prompt which included the command, bid question, bid specification, marking criteria & suggestions for improvement to ensure it meets the maximum marking criteria.

2. Search Prompt: An issue we've experienced with building first drafts is that using the expand function means that the AI would write an answer, which takes away the human element of bid writing somewhat. To overcome this, the search tool has been great at finding relevant text based on keywords I've built to associate with my pre-selected bid answers in my library; this allows me to pull the exact text from the sources for editing in my first drafts, which puts the control further back in my hands.

3. Library Management: The ability to create/sanitise your own bid library is much more efficient than looking for ideas/sources from a library with much wider subject matters. It feels like a constantly evolving product, in that the more relevant information you give it, the more relevant information you get back. While we do use a lot of Word/Excel/PDF files, what could help is an ability to pull stats from PowerBI dashboards we use to advance our libraries further. Review collected by and hosted on G2.com.

I do think some features need tweaking (in various levels depending on the experience) as they tend to add more time to the bid process rather than reduce time. For conciseness, I've listed the three features that I feel require the most improvement:

1. Research Assistant: It feels quite ineffective at pulling requested information from specific internet sources. For example, I asked it to give me a detailed list of all providers within an apprenticeship framework, even asking it to look at the URL that contained all this information, only to get a result saying that it was unable to do this. The result was right in front of it and it failed.

2. Gamma Review: Concerning my comments about Prompt/Pure Completion being an effective review tool, I think it does defeat the point of having Gamma Review as a feature. It has problems concerning word counts and lack of text boxes to provide the required information to ensure accurate review. While one could argue you could use the Extract/Summarise tool to make it fit the word count, you run the risk of compromising accuracy in your review process because some things just get omitted by default. In addition, while you can review answers against pre-defined criteria (e.g. innovation/evidence/robustness), bid writers will always defer to the scoring criteria given in each individual bid, making these pre-defined criteria seem inaccurate, although well-intentioned. To improve this, I think the system needs the following:

- Individual text boxes (without word counts) centered around the question, marking criteria, specification,

- Option for draft revision to ensure drafts hit maximum scoring criteria

3. Template Management/SSQ completion: Template Management could be a great tool for completing SSQ documents almost instantly, but because the process of exporting content into these templates is jarring as the template seems to take on a different form, it leaves some potential/time on the table. It needs to just input text based on question recognition identified within sources (e.g. library, datasets, Q&A workbooks), leave the template in its pre-uploaded form, and thats it, that should be all completed.

I'm probably suggesting a whole new feature here, but some way to upload an SSQ draft and populate it based on library sources (e.g. question recognition) would save time for future submissions. Review collected by and hosted on G2.com.