Plain old Google seems better to me

I bought the DevonThink and DevonAgent bundle a few months back, and I don’t see much value to the DevonAgent part.  Hopefully I am not using it correctly, and somebody will point me in the right direction.

For instance, I’m looking for the list of VGA cards compatiable with Linux.  I put the following into Google and DevonAgent.

linux compatiable vga card

First of all, Google notices that I misspelled compatible

Did you mean: linux compatible vga card

And the results set is reasonable, and I can get useful information from many of the results, and then start to re-phrase my query.

DevonAgent returns 1390 pages, broken into approx 25 catagories starting with Keyboard, Processor, Hardware on down to the last 3 being Looking, Supports, Control, in the digest mode.

Clicking on the various topics for a few minutes, and I’ve yet to see anything remotely close to my intent.  Lots of ads for computers for sale, with vga cards for example.

If anybody wants to refute my impression of DevonAgent, I’m all eyes.




my answer would be "no". I use google for quick search for a person - email adress or short information of the institute he (less often she) works.

But if I am collecting material for a certain topic DevonAgent works more precisely and presents the material in a way that lets me quickly judge whether I need an information or not - and whether I should archive it in DT or not.

I am still overwhelmed about the results.

Give it another try!


I experienced the same situation this week, and figured I lacked the knowledge to use the program effectively.

I needed information about several sites in Buffalo, NY. I edited my plug-in list in DA, but was never able to structure a search that would produce links to the Buffalo Zoo or Science Museum.

In desperation, I turned to Google only, and got everything I needed quickly. Why would the Buffalo Zoo site show up on a Google search, but fail to make the short list of hits from a DA search employing Google, AltaVista, Ask Jeeves, Wikipedia and Yahoo?

I know the program is smarter than I am. I need to learn how to talk with it, and I didn’t find the instructions I needed in the manual or the help file.

As an example, I selected DA’s Internat (fast scan) set and entered the following search term:

Buffalo AND (NY OR “New York”) AND (zoo OR “science museum”)

In seconds that yielded 73 hits, all of which but 2 or 3 might be useful. I quickly found material on teaching uses of the Zoo and Science Museum, read about a destructive riot by teenagers on a free admission day at the Zoo, budget problems, the origin of the Zoo, and much more.

What’s neat about DEVONagent is that one can scan the Digest area in various ways. Clicking on the keyword “Attractions” resulted in displays of a number of pages about the zoo and science museum. Or, in the Pages area, one can view the Web page listing for the search. Selecting one reveals a digest of the text contents.

You could select from Digest the text and URL that you may wish to store in your DEVONthink database. Or you could click on a Page item to open the Web page, and save it as a Web page into DEVONthink, or add selected text/images from the Web page to your DEVONthink database.

DEVONagent also offers the possibility of much deeper searches that would yield many more hits, or much more specific searches that would center about a very specific topical interest. Want to learn more about that Memorial Day riot at the Zoo? DEVONagent could probably find enough information to start you on a book about the event and its context in Buffalo history, demographics, economy and so on – and help you store your reference sources in your DEVONthink database.

We plan to provide more information about how to use DEVONagent, including examples and tips.

This forum is a good place to raise questions, and a good place for the user community to help each other, including How To and troubleshooting tips.


I used the Internet (fast search) set and entered this seach string:

linux AND VGA

In a few seconds, I got 67 hits (which is deceptive, as I discovered). One of the keywords under Digest was Card. So I clicked on that. There appeared to be hundred of descriptions of problems associated with VGA cards. I got the impression that “compatibiity” is a relative term, depending on the overall hardware and software flavors and configurations. Of the items I looked at, there seemed to be far more questions than answers. It seems that somewhere out there in Linux Land someone tries to keep a registry of compatible or “supported” video cards/drivers, but it also seems that the manufacturers may be reluctant to supply the information necessary to be listed.

The answer to your query may have been in my search results, but I don’t know enough about your machine to recognize it. Don’t recall seeing any ads, though there were card manufacturers mentioned. Also got the impression that if a card is supported under XFree86, that’s probably a good place to start.

Makes me glad I’m not building a computer. The video card in my 4-year-old TiBook may be slow and lacking in memory, but at least I’ve never had to worry about compatibility problems. :slight_smile:

DevonAgent has some interesting ideas, but it’s certainly something I’d use only when Google failed, especially since I can do four or five Google searches in the time it takes DA to do one.

It’s useful as a backup when Google fails, though.

To me they’re not even comparable. I used google for only the simplest of searches and I use DA for finding things that are most relevant for my research. It digs much deeper, to multiple levels, allows me to specifically add sites, allows all sorts of complex Boolean search parameters (I now use NEAR a lot), can ignore duplicates and do all kinds of other filtering, and then comes back with nice keyword based summaries of everything it finds. I can’t imagine how you could not like it. Google doesn’t come close imho except for the simplest of things.


Thanks, Bill. I attempted to reproduce your effort and broke through the brambles. I’m going to practice developing search terms and try to prove myself a worthy practitioner of this program, the breathtaking powers of which I can only imagine at this point.

I look forward to more discussions. Having someone like yourself monitoring these threads and offering guidance is worth more than all the printed manuals ever printed.

I agree with you for very simple searches, especially one-time searches. Example: To check for reviews of a digital camera, I might do a Google search for “FZ20” and quickly find links to several in-depth reviews of a nice Panasonic digital camera. Google did a great job for me! Perhaps that’s really all I need.

So why might I find DEVONagent useful?

SCENARIO: Suppose I’m planning to buy a digital camera, and I want to compare several cameras that have a long zoom length. A few digital cameras come with lenses that can zoom up to 10x or 12x. One of them is the Panasonic FZ20. I’d like to find some comprehensive review articles, and perhaps compare the FZ20 to other makes. I’d like to keep some or all of the search results for a whlle, and I may wish to store some of them in my DEVONthink database. Here’s what I would do:

[1] Select DA’s Internet (fast scan) search set and enter the search strin “FZ20” (without quotes). Press Start.

Observation: There were 62 results. But DA took longer than Google to display results. Why? Because DA has scanned and cached the text of each result, and Digest summaries have been tagged with topic keywords.

[2] Click on Digest and explore the results. The Digest lists several topicss (which may or may not look arbitrary). Since I’m looking for reviews, click on the Reviews topic. Several of the best digital camera review sites have been picked up, and I can look at digests of the text content of those sites.

The Topics list includes Panasonic (which makes the FZ20), Olympus, Kodak, Canon and Sony. It turns out that the other manufacturers make comparable long-zoom digital cameras, and model names of comparable cameras have been picked up in this search. That gives me good leads to check them out.

Clicking on the Image topic displays text that contains that term. Scanning through this, I can quickly find that image quality is good, there’s an effective image-stabilization feature, and so on. I see that sample images are available.

There’s no “Lens” topic. But I want to know more about the lens used in this camera. There’s a search field at the bottom of the Topics column. Type “lens” (without quotes), and I see that there’s lots of information about the lens, and it’s a good one.

Note: The text summaries and Digest display and search features will continue to be available even if I go offline.

[3] I decide to keep some of the search results on my computer for future reference. I don’t need all 62 results, so I click on “Pages”, hold down the Command key and select items, including the major review articles and some user review notes. In this case, I highlighted 12 results. Now, pressing the “Add” icon in the toolbar creates a new archive named “FZ20” and saves the source code of the selected pages.

Now, at any time, I can open Archive. If I click on the “FZ20” group, the 12 saved items are listed. If one of these items is selected, a summary of its text content is displayed.

Double-clicking on a listed item opens it in DA’s Web browser. If I decide to copy a page (or a selection of the page) into DEVONthink, that’s easy. Either use the contextual menu option Add Page to DEVONthink, or go to the menubar and select Data > Add to DEVONthink / HTML Page (there are similar procedures for capturing selections of text/images).

Bottom line: DEVONagent found useful information with a simple search (which Google can do). It also provided services that Google can’t do, including scanning and summarizing the text content of all the search hits, organizing result summaries under topics, and providing for Digest filter searches such as, in the example, “lens”. Some or all of the results can be stored in a DEVONagent archive with a single click, or moved to DEVONthink with a single click.

That’s just the beginning. DEVONagent gives the user the abiiity to:

  • Use complex Boolean search operators

  • Choose from or design search sets for specific sources of data

  • Drill down through multiple layers of pages to find information

  • Schedule follow-up searches that show only new results (for the past week, for example)

  • More (is anyone using scanner setting for images, for example?)

Many DEVONagent users are using DA search sets with search strategies that meet their needs. How about posting your DA tips and tricks for the edification of nerwbies (and me)? How do you use DEVONagent? Start posting your experience here, and we’ll develop an area of the Wiki site for How To and Tips topics (with attribution, of course). :slight_smile:

That’s a very useful example, Bill - I ‘get’ more of the ideas now! You’ve inspired me to go and read some more of the manual!

FYI, some of the things that affected my view initially were:

  • For most of what I do, knowledge of the individual words (in the digest) is not too helpful. This may be the key reason why DA hasn’t immediately jumped out as useful for me - I haven’t gained much from the Digest yet. If it could show two- or three-word phrases that occurred with the same frequency, that would be very useful (and harder to implement, I realise!). If the majority of uses of the word ‘Apple’ were actually in the context of ‘Apple Pie’ it would be interesting for that phrase to appear in the digest. Or the user could add selected phrases…

  • When you select a word in the digest, and then do a search, it’s not clear whether you’re searching just the pages with the selected word, or all of them. I think it’s latter, but the word is still selected… There’s a similar issue using the search box in the Archive window. Are you just searching the selected archive?

  • When viewing page summaries in the digest, it would be great if there were a way to add an individual one to the archive. I often see something I think interesting and then have to go back to Pages to try and find it before adding it to the archive.

  • It wasn’t clear to me whether I could refine an existing search without re-downloading all the pages. I guess the search field at the bottom does that.

  • My understanding is that pages are cached (by default) until the program exits. It would be nice if, after trying a new search which is less successful, I could return to an earlier search without re-loading the pages. I guess I can do that as long as I’m disciplined and always start every new search in a new window.

Anyway - many thanks! This kind of ‘example use’ is a great way to get people to appreciate some of the value.